fix(copyright): detect open-ended year ranges ending in "now" (#1222)

A copyright line whose year range ends in "now" (e.g.
"Copyright (c) 2013-NOW Jinzhu <wosmvp@gmail.com>") was dropped
entirely: the "2013-NOW" token failed every year pattern and tagged
as a common noun, so the grammar never built a copyright tree.

Extend the existing year-dash-present POS pattern (derived from
ScanCode's _YEAR_DASH_PRESENT) to also accept "now", matched
case-insensitively since it is commonly uppercase in the wild. The
match stays anchored to a year in the same token, so standalone
"now" and unrelated attached tokens like "2013-final" are not
treated as years.

Refs aboutcode-org/scancode-toolkit#5220

Signed-off-by: Maxim Stykow <maxim.stykow@gmail.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: mstykow

src/copyright/detector/tests.rs

@@ -81,6 +81,28 @@ fn test_standalone_c_holder_year_range_with_trailing_period_is_extracted() {
     );
 }
 
+#[test]
+fn test_open_ended_now_year_range_is_extracted() {
+    // scancode-toolkit#5220: an open-ended year range ending in "NOW" must not
+    // drop the entire copyright statement.
+    let input = "Copyright (c) 2013-NOW Jinzhu <wosmvp@gmail.com>";
+
+    let (copyrights, holders, _authors) = detect_copyrights_from_text(input);
+
+    assert!(
+        copyrights
+            .iter()
+            .any(|c| c.copyright == "Copyright (c) 2013-NOW Jinzhu <wosmvp@gmail.com>"),
+        "copyrights: {:?}",
+        copyrights.iter().map(|c| &c.copyright).collect::<Vec<_>>()
+    );
+    assert!(
+        holders.iter().any(|h| h.holder == "Jinzhu"),
+        "holders: {:?}",
+        holders.iter().map(|h| &h.holder).collect::<Vec<_>>()
+    );
+}
+
 #[test]
 fn test_obfuscated_angle_email_is_kept_in_copyright() {
     let input = "(C)opyright MMIV-MMV Anselm R. Garbe <garbeam at gmail dot com>";

src/copyright/patterns.rs

@@ -93,7 +93,10 @@ fn build_pattern_list() -> Vec<(String, PosTag)> {
     let year_short_punct = &format!("{}{}", year_short, punct);
     let year_or_year_year = &format!("({}|{})", year_punct, year_year_punct);
     let year_then_short = &format!("({}({})*)", year_or_year_year, year_short_punct);
-    let year_dash_present = &format!(r"{}[\-~]? ?[Pp]resent\.?,?", year);
+    // Provenant extension: also accept "now" as an open-ended range end, e.g.
+    // "2013-NOW". ScanCode only handles "present" here; "now" is commonly
+    // uppercase in the wild, so match it case-insensitively.
+    let year_dash_present = &format!(r"{}[\-~]? ?([Pp]resent|(?i:now))\.?,?", year);
 
     let mut patterns: Vec<(String, PosTag)> = Vec::with_capacity(1200);
 

src/copyright/patterns_test.rs

@@ -195,6 +195,14 @@ fn test_year_ranges() {
     assert_eq!(p.match_token("2010-202x"), PosTag::Yr);
     assert_eq!(p.match_token("1999,2000"), PosTag::Yr);
     assert_eq!(p.match_token("2020-present"), PosTag::Yr);
+    // Open-ended range ending in "now" (scancode-toolkit#5220), including the
+    // common uppercase form.
+    assert_eq!(p.match_token("2013-NOW"), PosTag::Yr);
+    assert_eq!(p.match_token("2013-now"), PosTag::Yr);
+    assert_eq!(p.match_token("2013now"), PosTag::Yr);
+    // A year attached to an unrelated word must NOT become a year range.
+    assert_ne!(p.match_token("2013-final"), PosTag::Yr);
+    assert_ne!(p.match_token("2020-Jinzhu"), PosTag::Yr);
 }
 
 #[test]